Crisis Time for Linux?

It was a rough ride for Linux in 2001. Corel scaled back its ambitious plans to take the desktop by storm, the once-promising Nautilus GUI project fell into disarray after Eazel, the company behind it, ceased operations in May after failing to obtain a second round of financing. Dell declared Linux too technical to ever make a significant dent in the desktop OS market. VA Linux, a leading provider of software and services for Linux OS users, declared a huge loss on May 22 and said it was unlikely to meet its stated goal of profitability by October, 2002. And then, days later, the CEO of MandrakeSoft, the leading Linux vendor, resigned. What's going on in Penguinland?

Perhaps the larger question is the one debated in a spirited exchange between Tim O'Reilly and Salon columnist Andrew Leonard, who asks "Just what role did the bubble economy of the '90s play in free software's march to prominence?" O'Reilly, an activist for Internet standards and for open source software, challenges the idea that open source business models have to look like proprietary software business models or they somehow don't count. He asks the corollary questions: Will the Internet go away because the dot com bubble burst? Did the personal computer go away because most of the startups that fueled the PC boom in the early 80s went belly up? The failure, he argues, of individual companies--even lots of them--has very little to say about the success or failure of technology. "In fact," says O'Reilly, "you can argue that the failure of many companies is required, if the industry is to progress. If there aren't enough failures, there's not enough innovation."

Andrew Leonard, in a gloomy editorial on the failure of Eazel, which burned through 16 million (US) dollars its first year before going belly up, wonders if free software is still relevant to the contemporary software marketplace. To be sure, some of the developers behind Eazel, including Andy Hertzfeld, who was part of the original team that developed the Macintosh, were atypical. Hertzfeld says "I've been compensated enough from my previous adventures so I can work on free software without pay indefinitely." Must be nice, but it's hard to justify as a business venture.

Part of the problem is the fact that, despite its progress, Linux is still immature in many areas. Sure, there's lots of free software, but it is uneven in quality and all too much of it lacks the kind of graphical consistency that made the Mac OS a success. This is why the demise of Eazel is doubly disappointing. Although Hertzfeld vows he will continue to host what's left of the Eazel website and Nautilus codebase "for the foreseeable future," the Linux community still suffers from all too many underdeveloped graphical user interfaces. Both KDE and Gnome ship with virtually all major Linux distributions, and there are at least a dozen other GUIs in wide distribution as well. Software developers who try to make sure their packages run under multiple graphical user interfaces do so knowing that they cannot strongly tie the code to a feature available only to one GUI. The developers are hamstrung by the code fragmentation that has left the Linux community divided into factions: "Red Hat compatible," "Debian compatible," Gnome vs. KDE, and what have you.

Ironically, what would probably be best for Linux as a business move is close to anathema for its user community and runs counter to virtually every tenet of the open source software movement's ideology: consolidate all the various incarnations of Linux, standardize on one graphical interface, one set of standard commands and go with it. Developers would have a vastly larger customer base to sell to, and Debian Linux users facing Mandrake Linux for the first time wouldn't have to wonder why the graphical interface is radically different, the programs they use aren't present, and even basic system functions such as hardware device detection and software installations are handled differently.

Indeed, market forces may be nudging Linux in the direction of consolidation already.

Mandrake Linux and Red Hat Linux, the two best-known Linux distributions, are already very similar in their code base and underlying operating system kernel. Both have ballooned to more than 1200MB of standard files, requiring multi-CD installations for basic operation if you choose to install both Gnome and KDE. Both support the "RPM" Red Hat Package Manager method of downloading and installing software updates. And, perhaps most importantly, the two are the leaders in the group of only three Linux distributions that really matter much anymore.

Figures released in May by market research organizations PC Data and LinuxGram show Mandrake Linux heading the list of U.S. retail Linux sales in the first quarter of 2001.

Those figures showed that Mandrake Linux represented 33.8 percent of Linux retail sales in the quarter, with Red Hat Linux at 30.7 percent, SuSE at 23.8 percent, FreeBSD at 5.6 percent, Caldera at 2.5 percent, Corel with 2.3 percent and Turbolinux at 1.2 percent. (It should be noted that MandrakeSoft released a new distribution during this period, a move that likely boosted its numbers over rival Red Hat, which has since followed with its own update.)

To put all these numbers into perspective, they all add up to less than 2 percent of the desktop market in total.

Are We There Yet?
PC Buyer's Guide tested both of the Mandrake 8.0 and Red Hat 7.1 distributions and found that, while multimedia support and plug-and-play detection of devices has improved, rough edges persist.

Our tests of Mandrake 8, for example, were interrupted twice by system freezes while attempting to configure network cards and the system died -- consistently -- with a catastrophic "black screen of death" failure when attempting to shut down. Later, a Red Hat installation on the same hardware failed to properly detect a TNT graphics card and left the system with a completely black screen. (As a point of interest, the same machine loaded up and ran Windows XP beta 2 without a hitch.) In all cases, the hardware was on the "supported" list, but uneven device drivers and quality control problems made configuration a hit-and-miss proposition at best.

It's no wonder that Rick Thwaites, Dell's European head of marketing, declares Linux "a fundamentally technical operating system" and proclaims it unsuitable for the novice user. Until a Linux desktop setup is as easy as that offered by the latest operating systems from Microsoft and Apple, Linux is likely to remain a favorite of system administrators, webmasters and supergeeks -- and the curious with plenty of time on their hands and a penchant for solving puzzles. Even more than 10 years after its introduction, Linux remains primarily, as its originator once described it, "a program for hackers by a hacker."

For Further Reading:

• ZDNet: Exec shakeup rocks Linux specialist Mandrakesoft
• CNET: Dell- Linux too tech for the desktop
• CNET: VA Linux's losses grow fourfold
• CNET: Consumer-Linux company Eazel closes
• Google:  Linus Torvalds' Usenet announcement of his "pet project"